Continuous Glucose Monitoring in Practice: When the Fasting Labs Do Not Tell the Whole Story

MG is a composite case built from a woman in her mid-forties who had been seeing the same primary care provider for years and had never been told anything more alarming than “keep an eye on it.” She was active, ate what she considered a reasonably clean diet, and did not fit the visual profile that either she or her prior clinicians associated with insulin resistance. What brought her into a functional medicine setting was not a dramatic lab flag. It was a pattern she could feel but could not explain: afternoon energy crashes that had become more consistent over the past year, a stubborn resistance to body composition changes despite consistent training, and a quiet frustration that every annual panel came back with the same verbal reassurance, borderline, keep doing what you are doing, even as she suspected something was shifting underneath the surface.

That gap, between how a patient feels and what a standard panel says, is often where the most useful case studies begin. MG’s chart had years of data sitting in it already. The story was there. It simply had not been read as a trajectory.

The Lab Trajectory: A Slow Drift

Pulled in isolation, none of MG’s early labs would have prompted concern. Several years back, her fasting glucose moved in the high 80s to high 90s, her A1c sat at 5.6 to 5.7 percent, and her fasting insulin was a very low 3.0 uIU/mL, a number that on its own reads as excellent insulin sensitivity. Her lipid panel from that period was suboptimal with a total cholesterol in the 190s to 220s, an Apo B in the low 100s, HDL comfortably in the mid to high 50s. This is the panel that, seen in a single visit, may not generate further workup. 

The picture began to shift gradually enough that it was easy to miss from one annual visit to the next. Her fasting glucose crossed into the low 100s. Her A1c ticked up to 6.0 percent. Her fasting insulin, while still within a broadly normal reference range, had climbed from 3.0 to 4.9 uIU/mL. At the same time, her lipid panel started moving in a direction that tends to travel with early insulin resistance, even before glucose numbers become dramatic. Total cholesterol climbed into the low 240s, triglycerides rose, and her Apo B, tracked across several consecutive draws, moved steadily upward year over year, from the low 100s into the 130s and eventually the low 140s.

By her most recent visit, the trajectory was unmistakable once it was laid out chronologically rather than read one panel at a time. Fasting glucose was 107 mg/dL. A1c had held at 6.0 percent. Fasting insulin was 5.4 uIU/mL, and for the first time, a C-peptide was drawn alongside it, coming back at 1.89 ng/mL, itself unremarkable on its own. Her most recent lipid panel showed total cholesterol at 238 mg/dL, HDL had fallen slightly to 49 mg/dL, triglycerides were up to 181 mg/dL, LDL was 157 mg/dL, and Apo B had reached 137 mg/dL. A coronary artery calcium score from earlier in her workup had come back at zero, and her Lp(a) was low and stable. Her omega-3 index, checked around the same time, came back low.

This is the moment in a chart review where a practitioner has to decide what to do with a set of numbers that are individually borderline but collectively directional. None of MG’s values, taken one at a time, may have triggered an automatic referral or a pharmaceutical conversation by her conventional practitioner. Through a Functional Medicine lens, looking at this as a trajectory, spanning several years, with the direction of travel across glucose, A1c, insulin, and every lipid marker tracked and moving in the wrong direction. We would have likely intervened earlier but we are here to help now!

Composite Lab Trajectory

Clinical Pearl: A coronary artery calcium score of zero is reassuring for existing calcified plaque burden, but it does not rule out active glycemic and lipid processes that are actively driving future risk. Do not let a favorable CAC score close the door on further glycemic investigation when the trendlines on A1c and Apo B are both moving upward.

The Discordance Problem: Why Normal Insulin Does Not Mean Normal Glucose Handling

Look closely at that trajectory and one pairing stands out. Across two separate draws roughly eight months apart, MG’s fasting insulin and C-peptide sat comfortably within normal limits. If a clinician stopped at those two values alone, the natural conclusion would be that beta cell function and insulin sensitivity were intact. Everything else in the panel argued otherwise.

Fasting insulin and C-peptide are single time point measurements. They tell you what beta cell output looked like at one moment, under fasting conditions, and nothing about what happens across the other sixteen or more waking hours of a day. A1c, meanwhile, is an average. It can mask significant postprandial spikes if a patient’s overnight and fasting glucose stays low enough to bring the average back down. It’s also not always the most reliable marker as there are other variables that can impact this number (we explain this more in our Practitioner Training Program curriculum). This is the exact blind spot that a single glucometer reading or a quarterly lab draw cannot resolve, and it is the reason we moved MG from static testing to continuous glucose monitoring rather than simply repeating the same labs and waiting for a more dramatic abnormality to appear.

CGM in Practice: When and Why We Order It

In our practice, continuous glucose monitoring is not reserved for patients who are already diagnosed with prediabetes or diabetes. We order a CGM for any patient where we suspect blood sugar dysregulation is present but not yet fully visible in conventional labs. This includes patients with:

Discordant conventional labs, such as MG’s pattern of normal fasting insulin alongside a rising A1c.

Strong clinical suspicion despite normal labs, including patients reporting energy crashes, sugar cravings, afternoon fatigue, or difficulty with body composition changes.

A personal or family history that raises concern, even when current labs are within range.

Patients are already engaged in active metabolic intervention, where we want objective feedback on how diet, movement, or supplementation changes are actually affecting glucose in real time, rather than waiting for the next lab draw.

We also use glucometers or glucose meters in practice, and they remain a valuable and far less expensive tool. A glucometer can confirm a fasting value or a single postprandial check with reasonable accuracy. What it cannot do though, is capture the shape of the curve. It cannot show what happens overnight, it cannot capture the timing and magnitude of a spike relative to a specific meal, and it depends entirely on the patient remembering to test at the right moments. CGM removes that burden and gives us continuous data across day and night, which is where the discordant cases like MG’s tend to reveal themselves.

Brand Landscape

Several CGM options are available to practitioners and patients today, and the landscape has split into two distinct tiers as more devices become available without a prescription. 

Abbott FreeStyle Libre and Dexcom’s prescription line remain the two dominant systems on the insurance side, and that coverage is still overwhelmingly tied to insulin use. 

On the direct-to-consumer side, Dexcom’s Stelo was the first sensor to reach the market without a prescription, built for people with type 2 diabetes who are not on insulin or who simply want general glucose feedback. Abbott has since answered with two separate OTC products of its own, Libre Rio, aimed squarely at the same non-insulin type 2 population Stelo targets, and Lingo, a more general wellness oriented sensor built on the same Libre technology but marketed to a broader consumer audience, including people without diabetes.

Stelo is the sensor we most commonly use in our own practice on a cash pay basis. We typically order two sensors, which provides roughly 15 days each of continuous data. This is generally enough time to capture a normal eating pattern along with a few intentionally varied days, without requiring the patient to commit to a longer and more expensive wear period.

Clinical Pearl: Insurance coverage for CGM remains largely tied to insulin use. Most patients presenting with the kind of subclinical, discordant pattern described here will not meet insurance criteria for coverage, which is why cash pay access to a device like Stelo has become a practical and reliable option for this population.

Advantages and Nuances of CGM Data

The advantages of CGM over intermittent glucometer checks or static labs are substantial. Continuous data reveals postprandial peak magnitude and the time it takes glucose to return to baseline after a meal, information a single 45 minute post-lunch check simply cannot capture in full. It reveals overnight patterns, including whether glucose is dropping and rebounding in ways that may point to counter-regulatory hormone activity or poor sleep quality. It allows a patient to see, in real time, how a specific food, a specific stressor, or a poor night of sleep shifts their glucose curve, which tends to build far more durable behavior change than a lecture about carbohydrate quality ever will.

There are nuances to interpretation that matter clinically. Consumer grade CGM sensors are generally less precise at the low end of the glucose range than laboratory grade testing, and a single day of atypical eating or unusual stress can distort the overall picture if a practitioner is not looking at the data across the full wear period. Sensor accuracy can also be affected by compression during sleep, dehydration, and certain medications. For these reasons, CGM data is best interpreted as a pattern recognition tool layered on top of, not a replacement for, conventional labs.

Findings from MG’s CGM Data

MG wore her first Stello sensor for fourteen days, with 78 percent sensor active time, more than enough for a reliable read on her patterns. At the aggregate level, the report was reassuring. 

Her average glucose was 104 mg/dL, her glucose management indicator, the CGM’s estimate of A1c based on mean glucose, was 5.8 percent, and she spent 100 percent of her wear time within the 70 to 180 mg/dL target range, with zero percent of readings flagged as high, low, very high, or very low. One note is that this “target range” is set up by Dexcom, we have different target ranges for our patients of 70 to 140 mg/dl.

Glucose variability, measured as coefficient of variation, came in at 11.6 percent, well under the 36 percent goal threshold. The device’s own automated pattern insight tool returned “no adverse glucose patterns detected.” Taken in isolation, this is the kind of report that could tempt a practitioner to close the glycemic chapter of the workup entirely.

The value of CGM in a case like this is not only in catching dramatic dysfunction. It is in what the aggregate numbers do not show. Looking at the day by day detail rather than the fourteen day summary, several specific and reproducible patterns emerged.

Meal specific postprandial spikes. MG’s day to day glucose sat in a tight, flat band, generally between 90 and 120 mg/dL, with only a handful of days breaking meaningfully above that. Every one of those breaks lined up with a specific, identifiable meal. A high carbohydrate dinner of a pretzel, ramen, and a non-alcoholic beer pushed her into the 140 to 155 mg/dL range. A lunch pairing broccoli soup with a slice of pizza produced her single highest reading of the entire fourteen day report, 160 mg/dL. A lentil soup and orange lunch produced a smaller but still notable rise to 138 mg/dL, which she flagged herself in her log, wondering aloud whether it was the fruit. By contrast, a burger eaten with only half the bun, a small portion of fries, an apple, and almond butter, a meal with meaningfully more fat, fiber, and protein woven through it, produced no meaningful excursion at all, even eaten later in the evening.

A responsive, favorable pattern with movement. On the night of the pretzel, ramen, and beer dinner, MG logged a fifteen minute brisk walk after the meal and noted that it brought her glucose back down to normal. The data confirmed it. This is a genuinely useful finding to hand back to a patient, since it turns an abstract recommendation to “move after meals” into a demonstrated, personal, and repeatable result she generated herself.

A data quality note worth flagging, not a nocturnal dip. You may notice an isolated low reading of 66 mg/dL around 6 am. This may seem alarming at first and often causes concerns for patients. This patient in particular thought she needed to consume honey, yogurt and granola to “get her blood sugar back up” In this case, the reading immediately before and the reading immediately after sit solidly in the normal range, with no gradual descent and no gradual recovery, just a single point that drops out and returns on the very next check. A genuine overcorrection low driven by a rapidly absorbed sugar source would be expected to trend down and climb back over several consecutive readings, not snap back instantly. This pattern is far more consistent with a transient sensor compression or connectivity artifact, a known limitation of interstitial CGM sensors during sleep, than with an actual hypoglycemic event. It is worth warning patients about this prior to collecting and also showing a patient directly on their own data, both to avoid manufacturing concern over a reading that was likely noise, and to teach the broader skill of reading a CGM report for shape rather than reacting to any single number in isolation.

A second layer of discordance. MG’s CGM derived GMI, 5.8 percent, came in meaningfully lower than her most recent laboratory A1c of 6.0 percent. This is worth exploring, because it is easy to assume one of the two numbers is simply wrong. GMI and laboratory A1c can diverge for reasons that have nothing to do with average glucose exposure, most commonly differences in red blood cell lifespan and turnover. MG’s CBC from the same testing window showed a mildly elevated hematocrit, which is a plausible contributor to a laboratory A1c running slightly ahead of what her actual glucose exposure would predict. This does not invalidate either number. It is a reminder that A1c is a proxy built on an assumption of average red blood cell turnover, and when that assumption does not hold precisely, CGM becomes the more direct read on true glycemic exposure.

Taken together, this is not the picture of a patient in overt glycemic crisis. It is the picture of a patient with excellent baseline glucose control who nonetheless has clear, reproducible, food specific spikes tied to lower quality carbohydrate meals, a pair of isolated low readings that read as sensor artifact rather than true nocturnal instability, and a laboratory A1c that is likely running slightly hot relative to her actual glucose exposure. None of this shows up on a fasting panel. The postprandial spikes are directly actionable, and they plausibly connect to her upward trending Apo B and triglycerides, since repeated postprandial excursions, even brief ones that do not move a fourteen day average, are enough over months and years to drive hepatic triglyceride production and small dense LDL particle formation

Treatment Sequencing

Nutrient and botanical support. Because MG’s aggregate glucose control was already strong, the nutrient strategy was built to target the specific mechanisms her CGM data pointed to rather than a generic antidiabetic stack. Berberine was introduced primarily for its lipid modulating effects alongside its role in glucose uptake, since her central concern was rising Apo B and triglycerides against a background of otherwise favorable time in range. Magnesium was prioritized as well, both for its broader role in glucose metabolism and because deficiency is common and frequently overlooked in patients with this presentation. MG had already begun experimenting with inositol on her own during her first sensor wear, noting it in her log the same evening as her post-dinner walk, and we formalized this into her plan given its supportive role in insulin signaling. Specific dosing protocols and the full nutrient sequencing we use for this pattern are covered in depth inside Adapt Functional Medicine Practitioner Training and Certification Program. 

Dietary sequencing targeted to her own data. Rather than a broad, unfocused dietary overhaul, we used MG’s own CGM report to identify the exact meals responsible for her excursions and built the plan around them. The pretzel, ramen, and beer style dinner and the pizza and soup lunch were both refined carbohydrate heavy and light on protein and fat, and both produced her largest spikes. The burger meal, eaten with fat, fiber, and protein woven throughout, produced no excursion at all despite being eaten later in the evening. We used this contrast directly with MG to reinforce the principle of pairing rather than restriction, anchoring carbohydrate containing meals with protein, fiber, and fat rather than eliminating any food category outright and eating foods in an order that helps with glucose uptake and metabolism. 

Reinforcing post-meal movement. MG had already discovered, on her own, that a fifteen minute walk after a high carbohydrate dinner brought her glucose back to baseline. We formalized this into a standing recommendation for any meal she anticipates will be carbohydrate heavy, giving her a specific, low effort tool she had already proven works for her own physiology.

Sleep and circadian considerations. Given the connection between sleep quality and next day glucose handling, we addressed her sleep hygiene directly, including consistency of sleep and wake timing and evening light exposure, both of which influence next day insulin sensitivity independent of diet.

Monitoring and retesting. MG’s plan includes a second CGM wear period after several months of intervention to objectively confirm a reduction in the size and frequency of her meal specific spikes and the resolution of her overnight dip pattern, alongside a repeat conventional panel including A1c, fasting insulin, hematocrit, and the full lipid panel with Apo B. The repeat hematocrit will help clarify whether her laboratory A1c continues to run ahead of her CGM derived GMI, and the repeat lipid panel will show whether the reduction in postprandial glucose exposure is translating into an improvement in Apo B and triglycerides as well.

Clinical Takeaways

Discordance between fasting insulin, fasting glucose, and A1c is a signal, not a reassurance. When these values do not agree, the fasting labs are frequently missing the postprandial and overnight story, and continuous glucose monitoring is often the most direct way to see it. CGM is not only a tool for patients already diagnosed with dysglycemia. Used proactively in patients with discordant labs, unexplained lipid trends, or strong clinical suspicion, it becomes both a diagnostic instrument and a behavior change tool that a static lab draw simply cannot replicate.

Cases like MG’s, where the labs disagree with each other and the story only becomes clear once you add the right diagnostic layer, are exactly what we teach practitioners to recognize and act on inside Adapt Practitioner Certification and Fellowship Program. If you want to build the clinical reasoning skills to catch these patterns earlier in your own patients, this is where that training happens.

The next cohort of the Practitioner Training Program begins October 24, and we are currently accepting deposits to hold your seat. Space in each cohort is limited, and a deposit locks in your spot ahead of the general enrollment window. Reserve your seat with a deposit today.